Vision Transformer for Cell Tumor Image Classification

With the successful application of deep learning models in medical image classification tasks in recent years, especially in tumor recognition, we explore the application of Vision Transformer (ViT) model in cell tumor image classification. In this paper, we improve the performance of the pre-trained Vision Transformer (ViT) model on this task by fine-tuning it on the cell tumor image dataset, and compare its performance with other state-of-the-art tumor image classification models, including convolutional neural networks (CNNs). Experimental results show that compared with the convolutional model, the Vision Transformer(ViT) model has better globation ability and interpretability. In addition, we conduct a detailed analysis of the attention mechanism of the Vision Transformer (ViT) model, which provides important insights into the model’s decision-making process. Therefore, our study shows great potential for the use of Vision Transformer (ViT) in medical image analysis tasks, especially in tumor classification.